Water suppression using a combination of hard and soft pulses

Abstract

Numerous methods are available for suppression of the Hz0 resonance in the NMR spectra of water-soluble compounds. These methods include presaturation (I, 2), excitation with a soft pulse (3) or a Redfield pulse (4), rapid scan correlation (5), binomial hard-pulse sequences (6-9), and methods that exploit the different relaxation characteristics of HZ0 and solute protons (10-13). In recent years, the binomial hard-pulse sequences and variations thereof (14, 15) have become most widely used. Morris and Smith (16) showed that the binomial sequences can be combined with a single nonselective 90” pulse to provide an excitation spectrum which has nearly maximal excitation over a wide frequency range except for a narrow region around the HZ0 resonance which has near-zero excitation. The phase distortions resulting from this type of excitation are severe which leads to problems when this type of composite pulse is used in phase-sensitive two-dimensional NMR experiments. Levitt and Roberts (I 7) very recently proposed a new hard-pulse sequence which has a much weaker phase dependence across the spectrum while still providing good water suppression. However, the remaining small phase distortions, which present no problem in 1D NMR, can present difficulties in 2D NMR spectra. For example, cross peaks in a NOESY spectrum are typically two orders of magnitude weaker than diagonal peaks and even small phase distortions of the diagonal resonances give rise to serious distortions in the 2D spectrum. A recently proposed echo scheme (18) yields pure absorptive resonances across the entire spectrum but gives poor excitation close to the water resonance. Here we demonstrate the use of a hard/soft pulse combination first suggested by Gupta and Redfield (I 9). In principle, this type of hard/soft pulse combination can solve both the excitation window and the phasing problems mentioned above. However, it should be noted that the hard/soft pulse combinations described below require very stable spectrometer hardware. Conceptually the method is extremely simple. With the carrier placed on the HZ0 resonance, a weak 90?, pulse of duration T (T = 5 ms) rotates the HZ0 magnetization from the z axis to the -y axis (Fig. la), while leaving all resonances more than about 50 Hz removed from the HZ0 resonance along the z axis. A subsequent nonselective 90,” pulse rotates the HZ0 magnetization back to the z axis and rotates the resonances